POLARIZATION RANDOM NOISE SUPPRESSION METHOD IN TWO-DIMENSIONAL FORCE SENSOR BASED ON RANDOM FOREST

- The technique of distributed optical fiber sensing is very effective and applied in many different industries. Because of its benefits, including a long sensing distance and good spatial resolution, Brillouin Optical Time Domain Analysis (BOTDA) has received a lot of interest. A method for reducing the polarization random noise of a two-dimensional force sensor based on random forest is proposed in order to lessen the restriction between long sensing distance and high spatial resolution and measurement accuracy. The random forest algorithm is presented first. The polarization correlation of Stimulated Brillouin Scattering (SBS) and its impact on Brillouin time domain analysis and sensing are then examined. By studying polarization random noise, a method to suppress polarization random noise based on random forest algorithm and Brillouin phase shift spectrum (BPS) is proposed. Following that, simulation studies are used to confirm the method's impact on enhancing sensing accuracy. Finally, an analysis is done of the impact of gray coded BOTDA system sensors on wavelet-based image threshold denoising technology. The findings demonstrate that while the wavelet threshold denoising method is excellent at denoising white noise, it is not successful at suppressing polarization random noise, which lowers the sensing accuracy. Polarization random noise is less of an issue with BPS than it is with Brillouin gain spectrum (BGS), which allows for a three-fold increase in sensing accuracy. Wavelet denoising has a denoising impact on Brillouin phase shift sensing data that is unquestionably superior to Brillouin gain, and the BFS fluctuation is unquestionably decreased after denoising. This paper offers some suggestions and guidelines for reducing polarization random noise in sensors.